Download Modeling of External Ear Acoustics for Insert Headphone Usage*

PAPERS
Modeling of External Ear Acoustics for Insert
Headphone Usage*
MARKO HIIPAKKA,1 AES Student Member, MIIKKA TIKANDER,2 AES Associate Member,
AND MATTI KARJALAINEN,1 AES Fellow
1
Department of Signal Processing and Acoustics, Aalto University, Espoo, Finland
2
Nokia Corporation, Helsinki, Finland
Editor’s Note: This Journal paper is a fully reviewed submission, which was awarded the Student Technical Papers Award
at the 126th Convention of the Audio Engineering Society in Munich, Germany, 2009 May 7–10. The student, Marko Hiipakka,
presented the paper and was honored at the Convention. The Editor hopes this will encourage future student submissions.
Although the acoustics of the external ear has been studied extensively for auralization and
hearing aids, the acoustical behavior of insert headphones is not equally well known. It is
laborious to measure the sound pressure at the eardrum for individual insert headphone users.
The present research focused on the effects of the outer ear on sound pressure at the eardrum
during insert headphone listening. The main factors of interest were the length of the canal
and the impedance of the eardrum. Ear canal simulators and a dummy head were constructed,
and measurements were also performed with human ear canals. The study was carried out
both with unblocked ear canals and when the canal entrance was blocked with an insert
earphone. Special insert headphones with in-ear microphones were constructed for this
purpose. Physics-based computational models were finally used to validate the approach.
The different methods used to investigate the pressure inside the ear canal gave similar and
accurate results. Hence the modeling techniques proved to be useful in estimating the pressure frequency responses at the eardrums of individual insert headphone users.
0 INTRODUCTION
0.1 Background and Previous Research
It is well known that the outer ear contributes to the
spectral shaping of the sounds we hear in everyday life
[1]. Humans have different ears and different ear canals,
hence the sound pressure responses at the eardrums of
human test subjects are not similarly distributed. In part
therefore, people perceive sounds differently.
In normal listening situations the whole outer ear contributes to the spectral shaping of sounds before they reach
the eardrum. An unblocked ear canal acts like a quarterwave resonator and hence amplifies the resonance frequencies. The locations of these resonance frequencies
depend mainly on the effective length of the ear canal.
The shape and size of the pinna and the curvature of the
ear canal also have an effect on the pressure frequency
response at the eardrum.
Insert-type earphones are increasingly popular when listening to music and together with mobile phones, yet their
behavior has not been studied thoroughly. The sound trans*Presented at the 126th Convention of the Audio Engineering Society, Munich, Germany, 2009 May 7–10; revised 2010
January 4.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
mission path from the insert earphone to the eardrum is
different than when listening to loudspeakers or acoustically
open headphones. The sound wave travels only, through the
ear canal, an ear canal that is suggestive of a half-wave
resonator. The half-wave resonance frequencies are pronounced at the eardrum, and the locations of these frequencies depend once again on the length of the ear canal. In
addition the overall structure of the ear canal has an effect on
the frequency response at the eardrum [2], [3]. Furthermore
the pressure chamber effect, the occlusion effect, and leakage are important factors with regard to insert earphones and
low-frequency reproduction. In this study, however, the
emphasis is on the frequency range of 1 to 15 kHz.
Several methods have been developed to measure the
ear canal or eardrum impedance of real ears [2], [4]–[17].
The coupling of headphones to the ear has been studied by
[18] among others, who concluded that there are significant differences in the frequency responses generated in
the ears of different individuals. Important results related
to the transfer characteristics of headphones in human ears
have been published [19], [20]. In the present work additional results related to the coupling of an insert earphone
to the ear canal are presented. The presented computational model and measurement methods are useful in the
design of insert earphones.
269
HIIPAKKA ET AL.
0.2 Goal of Study
The goal of our research was to learn what kinds of effects
the differences in the physical parameters of the outer
ears have on their frequency responses, both for open and
occluded canals. We aimed at constructing a physical dummy
head with adjustable ear canal features and having as accurate
a humanlike acoustical behavior as possible. Furthermore
we aimed at presenting physics-based computational models
for a mathematical understanding of the problem.
Simple tubelike adjustable ear canal simulators were found
useful in determining the effect of the ear canal length (for
details see Section 3). The frequencies of the ear canal resonances and antiresonance notches found through measurements were the same as those calculated with physics-based
formulas. The acoustical behavior of a simulator equipped
with a damped eardrum was much like that of the human
ear canal. In addition, a custom-designed dummy head used
in our measurements proved to be a fairly accurate model
of human peripheral hearing (details are given in Section 3).
We were hence able, with good accuracy, to study the effects
of the individual differences in the outer ear parameters.
Physics-based computational modeling with lumped
and distributed elements was also applied to open and
closed ear canal simulation (see Section 2). For a simplified open ear canal such a model is very accurate. In the
case of an insert earphone feeding the canal the main
problem is to estimate a good acoustic Thévenin equivalent for the earphone driver. With different model calibration techniques we managed to get close to the behavior
measured with physical simulators up to 20 kHz. These
modeling efforts help us to understand better how the
insert earphones work for individual listeners.
0.3 Overview of Simulators
Physical simulators, such as ear canal simulators and
dummy heads, have been used widely as substitutes of
human test subjects. The accuracy with which physical
simulators can imitate the behavior of the outer ear and
the ear canal depends on how well different physical
details have been taken into account.
Originally ear simulators were primarily targeted for
hearing-aid and audiometry headphone calibration. All
calibrators were designed to mimic the acoustical load of
real ears, or at least provide a load in the same range as a
real ear. Nowadays ear simulators are increasingly used
for headphone and mobile handset calibration as well. Ear
canal simulators for insert-type headphones and hearing
aids aim to offer a realistic representation of a human ear.
Ear canal simulators are designed to have standing wave
patterns similar to those of a real ear. The impedance is
also designed to mimic the real ear impedance. Occluded
ear simulators (commonly based on ANSI S3.25 [21] and
IEC 60711 [22]) define the acoustic transfer impedance at
the eardrum. The standards are defined in the frequency
range of 100 Hz to 10 kHz, thus the corresponding commercial simulators are calibrated in this frequency range
as well. ITU-T has published a recommendation (ITU-T
P.57 [23]) that extends the IEC60711 standard by combining different types of pinnae with the simulator. The ear
270
PAPERS
canals recommended in ITU-T P.57 are comprised of a
10-mm-long uniform tube, and as they are designed for
tight coupling of hearing aids, measurement results for
insert headphones yield an exaggerated bass response.
The exact time a manikin or artificial head for acoustic
research was presented for the first time cannot be determined exactly. The period from 1880 to 1930 can be
considered the initial phase of development of binaural
recording devices, with the inventions of Goehner, Jones,
Firestone, and Fletcher [24]. According to Firestone [25],
the first time a manikin was used as a recording device
was around 1928 to 1930. Firestone describes the use of a
manikin that imitated a human head made of wax and a
wooden torso. The microphones were Baldwin receivers,
placed where the ears would be. The dummy head was
used to investigate phase and intensity differences at the
microphones. During the years of 1960 to 1970 the development was rapid, with advances such as the head-related
transfer function (HRTF) and significant new knowledge
gained about the function of the pinnae [24].
1 MEASUREMENTS
The acoustic behavior of the outer ear and the ear canal
were studied through extensive measurements with ear
canal simulators, a dummy head, and human test subjects.
The goal was to learn how the outer ear behaves with and
without insert earphones. In order to measure the effects of
the differences in physical parameters of the outer ear a
variety of different ear canal simulators, different kinds of
artificial pinnae, and a dummy head were constructed. The
main focus of this study was on the effect of the length of
the ear canal and the effect of the eardrum impedance.
Therefore only the simulators constructed for these purposes are presented in this paper.
1.1 Earphone with Fitted In-Ear Microphone
(EFIM)
To measure the frequency responses of a blocked ear
canal, a special earphone was constructed. A Knowles
FG-23329 miniature microphone was fitted in front of
the transducer port of a Philips SHN2500 earphone, as
depicted in Fig. 1. When the acoustical behavior of a
blocked ear canal or ear canal simulator was studied, this
earphone with fitted in-ear microphone (EFIM) was
placed at the canal entrance, as depicted in Fig. 2. Impulse
Fig. 1. Pair of earphones with fitted in-ear microphones (EFIM).
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
response measurements were made using the exponential
sine sweep technique [26].
The transfer functions from the transducer to the in-ear
microphone in different situations (attached to a small
cavity, attached to a long acoustically terminated tube
with 9-mm diameter, and in free field) are depicted in
Fig. 3. The strong peak in the frequency response at 6
kHz is the earphone’s self-resonant frequency peak.
1.2 Adjustable Ear Canal Simulator (ADECS)
The main reason why ear canals are often modeled as
straight rigid wall tubes is related to the wavelength of
audible sound waves. The diameter of the ear canal is
smaller than the wavelength of the highest audible frequencies. The compliance of the stiff canal walls of adults
is negligible with respect to the middle-ear compliance
[10]. Hence a straight rigid wall tube acts as a good
starting point when building physical ear canal simulators.
To study the effect of the ear canal length an ear canal
simulator with adjustable canal length (ADECS) was
constructed (see Fig. 4). The “ear canal” is made of a hard
plastic tube with a diameter of 8.5 mm, which is slightly
larger than the average canal diameter of an adult male.
The cross-sectional area of the human ear canal varies
along the canal length, being approximately 35 mm2
(female) or 50 mm2 (male) [27].
The “eardrum” is a movable plastic piston, which
makes it possible to adjust the canal length from 0 to 39
mm. Fig. 5 shows a diagram of the ADECS. The structure
of the human eardrum as well as its conical shape and
inclination all affect the pressure frequency response in
the ear canal, especially at frequencies above 10 kHz
[28]. Inclined and conical versions of the artificial eardrum were also tested during this study, but only the
results for the perpendicular version are presented here.
Fig. 2. Diagram of measurement setup where EFIM is fitted to
undamped ADECS and ear canal length is adjusted by moving
eardrum piston.
MODELING OF EXTERNAL EAR ACOUSTICS
A miniature microphone was fitted in the center of the
piston. The position of the microphone along the canal is
manually adjustable. It can be located at the eardrum piston level or pushed out as far as 57 mm into the canal
toward (and outside of) the canal entrance.
An important characteristic of the human eardrum is that
it dampens the resonance peaks and antiresonance notches
caused by the ear canal. These notches and peaks are sharp
in a rigidly terminated tube, whereas they are smoother in
real ears. The human eardrum normally softens the mentioned notches and peaks by damping the sound wave that
reflects from the drum. The impedance of the eardrum
determines the magnitude of the reflecting wave at different
frequencies. To achieve a better analogy with the human
ear, a damped artificial eardrum was also fabricated. The
movable plastic eardrum piston used in the undamped
ADECS was replaced with a movable piston made of aluminum and consisting of the microphone and an adjustable
Helmholtz resonator. The resonator acts as a damper at the
eardrum as some of the sound energy inside the ear canal is
dissipated in the resonator. The volume of the resonator
cavity can be changed by sliding the back wall of the cavity.
Absorbing material was added inside the cavity to spread
the damping effect to a wider frequency range. A diagram
of the eardrum and the ear canal is shown in Fig. 5.
The goal was to achieve a behavior similar to the human
ear canal, especially when the simulator is used with insert
earphones. The best similarity with human ear canals for
frequencies between 1 and 10 kHz was achieved when the
Fig. 4. Adjustable ear canal simulator (ADECS) mounted to a
microphone stand. Eardrum microphone has been moved from
eardrum toward canal entrance.
Fig. 3. Transfer functions (frequency responses) from EFIM’s transducer to its microphone when attached to a small cavity, to a long
tube, and in free-field conditions.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
271
HIIPAKKA ET AL.
resonator resonance frequency was set to approximately
2 kHz, which is close to the first antiresonance notch with
the blocked canal (as depicted in Fig. 19). The amount of
absorbing material was also selected with care. An almost
similar approach was used in [29]. The damped eardrum
worked well as a practical solution, but further studies on
modeling the eardrum are nevertheless of great interest.
1.2.1 Unblocked ADECS Measurements
When studying the pressure frequency responses of the
ADECS, the simulator was mounted on a microphone stand
(as shown in Fig. 4) in an anechoic chamber and pointed
toward a loudspeaker at a distance of 2 m. The responses at
different points along the ear canal were measured using
the movable eardrum microphone of the simulator. The
ADECS was also used to study the behavior of the resonant
frequency peaks as the length of the “ear canal” changes.
The length of the ADECS canal was adjusted from 0 to
30 mm in 1-mm steps while the pressure frequency
responses were measured with the eardrum microphone.
Both the undamped and the damped eardrums were used
consecutively. The results are presented in Section 3.1.
1.2.2 Blocked ADECS Measurements
Compared to normal listening conditions with an
unblocked ear canal, the acoustic behavior of the ear canal
is very different when it is blocked with an insert earphone. When the signal is played with a loudspeaker all
the outer parts of the auditory system act together to form
the total transfer function. In a listening room, with the ear
canal left open, the acoustics of the room is also shaping
the frequency response of the system, that is, the transfer
function from loudspeaker to eardrum. When using an
insert earphone the effects of the head, shoulders, pinna,
and concha are absent and only the acoustic properties of
the ear canal are of significance. Nevertheless the outer
parts of the outer ear have a major role in shaping the way
we are used to hear external sound sources, which is
important, especially in directional hearing. These outer
parts should therefore not be completely forgotten when
contemplating the acoustics of the occluded ear canal.
During the research several measurements were performed with the EFIM mounted to an ear canal simulator,
as depicted in Fig. 2. An exponential sine sweep was
PAPERS
reproduced with the EFIM and recorded with its in-ear
microphone and the eardrum microphone of the ADECS.
Similar to the measurements with the open ear canal, the
effect of the canal length was studied, making good use of
the adjustable simulator. The length of the canal was
adjusted from 5 mm (minimum) to 40 mm in 1-mm steps.
A similar set of measurements was performed with the
damped eardrum. The canal length of the simulator was
adjusted as frequency responses were measured using the
eardrum microphone and the in-ear microphone. The
damped eardrum smoothens the antiresonance notch at
around 2 kHz (as can be seen from Fig. 20).
1.3 Dummy Head with Adjustable Ear Canals
(DADEC)
An ear canal simulator without pinna and head is only
suitable for occluded ear measurements. For measurements in the free field and a listening room, the next step
in our research was to build a complete dummy head. A
torso with the dimensions of an adult male human was
added to a manikin head. Acoustically realistic artificial
pinnae with different sizes and shapes were fabricated and
used with the dummy head. The ADECS with undamped
and damped eardrums was used as the ear canal. Both the
pinnae and the ear canals were interchangeable. The
dummy head with adjustable ear canals (DADEC) is
depicted in Fig. 6. In the present research the total length
Fig. 6. Dummy head with adjustable ear canals and interchangeable pinnae (DADEC).
Fig. 5. Diagram of ADECS. (a) Undamped eardrum. (b) Damped eardrum. Cross sections from front (left) and side (right).
272
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
of the ear canal of the DADEC is defined as the length of
the ear canal of the ear canal simulator plus a few millimeters since the artificial ear extends the ear canal, as
depicted in Fig. 7. The defined length of the open canal is
greater than that of the blocked canal.
To investigate responses at the eardrum of the dummy
head (DADEC) with the damped ear canal simulator
(ADECS) a set of measurements were performed. The
DADEC was placed at a distance of 1.8 m with an azimuth
angle of 30 in an anechoic chamber. The length of the ear
canal was adjusted in steps of 1 mm while the pressure
frequency responses were measured with the eardrum
microphone of the ADECS.
To study the effect of the pinna three different fabricated pinnae were attached to the DADEC with the
undamped ADECS as the ear canal in listening room conditions. In addition to the “normal” pinna used in most
measurements (see, for example, Fig. 17), two larger pinnae (“big” and “cuplike”) were used (see [30] for details).
1.4 Human Test Subjects
The pressure frequency responses of the blocked ear
canal were also measured with real human ears. The test
subjects were asked to place the EFIM in their ears, and the
responses were then captured using the earphone and its inear microphone. Eight test subjects (seven male and one
female) participated in these measurements, with both ears
being measured. The test subjects were between 25 and 35
years of age, and they had normal ears and normal hearing.
2 EAR CANAL MODELING
Computational modeling was applied to open and
closed ear canal cases in order to test our comprehension
of the phenomena involved in ear canal and earphone
acoustics. This was accomplished by comparing modeling
results against measured data from the physical simulators.
2.1 Unblocked Ear Canal Modeling
An open ear canal, closed at the other end by the eardrum, acts as a quarter-wavelength resonator. The resonance frequencies of a cylindrical tube open at one end,
such as the ear canal, are [31]
nc
fn ¼ (1)
8r
4 Lþ
3
MODELING OF EXTERNAL EAR ACOUSTICS
where n is an odd number (1, 3, 5, . . .), L is the length of
the tube, r is the radius of the tube, and c is the speed of
sound. The resonator produces only odd multiples of the
fundamental resonance frequency, which is an octave
lower than in a tube that is open at both ends.
An open ear canal simulator without a dummy head is
simply a terminated tube. It can be modeled as a system
composed of approximate eardrum impedance, a practically
lossless ear canal as an acoustic transmission line, and
an external pressure sound source with internal acoustic
impedance, equivalent to the radiation impedance of the
tube opening [32, ch. 5] due to the reciprocity principle
[33, ch. 7]. The equivalent circuit applied is shown in Fig. 8.
2.2 Blocked Ear Canal Modeling
To enable computational modeling of the interaction
between the earphone and the ear canal, a model of the
earpiece as an electroacoustic source is needed. Instead of
trying to derive an electromechanic-acoustic equivalent
circuit,1 it is enough to estimate by measurements a
Thévenin type source model with voltage-to-pressure
source term PS and acoustic source impedance ZS, as illustrated in Fig. 9. For a given acoustic load ZL, the pressure
response PL delivered by the source is
PL ¼
ZL
PS :
ZS þ ZL
(2)
In principle the easiest way to obtain the two unknowns is
to measure the open-circuited (ZL ¼ 1) pressure, equal to
PS, and the short-circuited (ZL ¼ 0) volume velocity QL ¼
PS/ZS, from which ZS is solved. In contrast to electric
circuits, a problem in acoustics is that both conditions
mentioned are difficult to obtain, and therefore other more
ideal loading conditions need to be applied.
There are a number of published methods to measure
and estimate the Thévenin source parameters. Many of the
methods have been developed for probes used to measure
the ear canal or eardrum impedance for audiology purposes
[2], [4]–[17]. The impedance probes in audiology typically
include a sound source driver fed through a thin vent and
a microphone through a probe tube, sensing pressure at
1
For headphone modeling in general, see [34] and [35].
Fig. 8. Circuit model for open ear canal. RS—source resistance;
LS—source inductance; PS—pressure source; ZW—canal wave
impedance; Ze—eardrum impedance.
Fig. 7. Simplified diagram of ear canal of DADEC. Total length
is length of ADECS extended by interchangeable pinna.
(a) Open ear canal. (b) Ear canal blocked with earphone.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
Fig. 9. Acoustic Thévenin equivalent circuit for earphone.
273
HIIPAKKA ET AL.
PAPERS
a short distance from the source radiation point. Responses
measured very close to the radiation source are not directly
applicable due to near-field effects [36].
A commonly used method of calibrating the impedance
probe is to load it with several hard-walled closed tubes or
cavities, for which there exists an analytically computable
impedance expression [6], [8], [11]. Using, for example, five
different lengths of tube loads, there will be five equations to
solve for two variables. This means an overdetermined set of
equations, leading to least-squares optimization of both the
source pressure term and the acoustic impedance [6], [8].
We tried many different methods to estimate a Thévenin
source model for the Philips earphone. The best results were
obtained when using several approximately resistive loads,
made of long tubes with different diameters, for which the
wave impedance is Zw ¼ rc/A, where A is the crosssectional area of the tube, r is the air density, and c is the
speed of sound. A tube length of about 2–3 m is long
enough when the back reflection from the open end is
removed by temporal windowing of the reflection function.
Five (M ¼ 5) different calibration loads Zi with diameters of 5, 6, 7, 8, and 10 mm were used to measure the
in-ear microphone pressure frequency responses Pi. The
source terms PS and ZS for the earphone driver were
solved in the least-squares sense from the overdetermined
set of equations
2
3
2
3
Z1 P1
P1 Z1
6 Z2 P2 7 6 P2 Z2 7
6
6
7 PS
7
(3)
¼ 6 .. 7
6
7
..
4
5 ZS
4 . 5
.
ZM PM
PM ZM
using the pseudoinverse function (pinv in MATLAB).
Fig. 10 shows the magnitude responses of the measured
pressures Pi, and Figs. 11 and 12 plot the magnitude behaviors of the source pressure PS and the impedance ZS, respectively. Due to the near-field effects of the sound source we
could not use the in-ear microphone of the earphone
(EFIM), but applied a separate probe microphone 7 mm
from the earphone outlet to measure the pressures Pi.
Fig. 10. Measured in-ear microphone pressure frequency responses in five resistive load tubes for Thévenin model estimation.
Fig. 11. Pressure magnitude of Thévenin model.
Fig. 12. Impedance magnitude of Thévenin model, normalized to wave impedance of 8-mm-diameter tube.
274
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
In the blocked ear canal model the acoustic load ZL
was composed of the eardrum impedance and a lossless
ear canal as an acoustic transmission line, similar to the
unblocked model.
3 RESULTS FROM MEASUREMENTS
AND MODELING
The measured and modeled pressure frequency
responses are presented in this section. Measurement
results are presented for the unblocked and blocked
ADECS, the DADEC with unblocked ear canal, and
human test subjects with blocked ear canals. Frequency
responses from the canal entrance are presented for the
ADECS and the human test subjects. Responses from the
eardrum are presented for the ADECS and the DADEC.
Comparisons between measured and computationally modeled pressure frequency responses are presented for the unblocked and blocked ADECS with the undamped eardrum.
3.1 Frequency Responses of Unblocked Ear
Canals
3.1.1 Frequency Responses of Unblocked
ADECS
The free-field transfer functions at the ADECS ear
canal entrance (with canal lengths of 20 and 25 mm)
are depicted in Fig. 13. The first two peaks of the resonance frequencies of the quarter-wave resonators are at 3
and 9 kHz for the 25-mm canal. In the frequency
domain the antiresonance notches follow directly after
the peaks. The frequencies at which these notches are
MODELING OF EXTERNAL EAR ACOUSTICS
located correspond to the distance between the measurement point and the eardrum, from where the reflected
sound wave arrives.
When a sound wave enters a tube that is closed at
the other end, the sound pressure has its maximum at
the closed end. Antiresonance notches do not exist at the
closed end, contrary to other parts of the tube. The
pressure peaks of the resonant frequencies of the quarter-wave resonator tube are, however, pronounced at the
closed end. Some examples of the measured responses
with the two different eardrums are shown in Fig. 14.
The effect of the damping is clear at the resonant frequencies of the ear canal, where the eardrum attenuates
the peak of the resonance frequency by approximately
10 dB.
3.1.2 Measured versus Modeled Frequency
Responses of Unblocked ADECS
Figs. 15 and 16 compare the measured and modeled
pressure frequency responses at the canal entrance and
the eardrum, respectively, for a 28-mm-long undamped
ADECS ear canal simulator of 8.5-mm diameter. The eardrum impedance used in the computational model was a
resistance R ¼ 200 MO. The radiation impedance was as
shown in Fig. 8. As can be seen, the measured and
modeled responses agree well.
3.1.3 Frequency Responses of Unblocked
DADEC
The free-field pressure frequency responses at the
eardrum of the dummy head (DADEC), with unblocked
Fig. 13. Measured pressure frequency responses at ear canal entrance of ADECS with two different ear canal lengths in free-field
conditions.
Fig. 14. Measured pressure frequency responses at eardrums of undamped and damped ADECS with different ear canal lengths in freefield conditions.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
275
HIIPAKKA ET AL.
PAPERS
ear canals, are depicted in Fig. 17. The second resonance at around 10 kHz has been attenuated compared to Fig. 14. This damping is probably due to the
improved impedance matching from the ear canal
through the concha and the pinna to the environment.
The concha and the pinna cause a hornlike expansion of
the wavefront.
The effect of the three different pinnae used in the
measurements (see Section 1.3), under listening-room
conditions, is depicted in Fig. 18. The overall differences
Fig. 15. Measured responses at ADECS canal entrance and eardrum for 28-mm canal length.
Fig. 16. Modeled responses at canal entrance and eardrum for 28-mm canal length.
Fig. 17. Frequency responses at eardrum of DADEC with various ear canal lengths in an anechoic chamber with azimuth angle of 30 .
Given length (*) is length of damped ADECS canal—not total length (see Sec. 1.3 and Fig. 7).
Fig. 18. Effect of pinna size and shape. Three different pinnae attached to DADEC. Responses are measured from undamped eardrum
in a listening room with 0 azimuth.
276
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
in the obtained frequency response curves are not striking
considering the exaggerated differences in the physical
sizes and shapes of the pinnae.
3.2 Frequency Responses of Blocked Ear Canals
3.2.1 Frequency Responses of Blocked ADECS
The frequency responses measured with the eardrum
microphone and with the in-ear microphone of the EFIM
earphone from the undamped ADECS with a canal length
of 25 mm are depicted in Fig. 19. The graph shows a sharp
antiresonance notch at approximately 2.2 kHz at the
earphone microphone (canal entrance). The first halfwave resonance, which is marked by an arrow, is approx-
MODELING OF EXTERNAL EAR ACOUSTICS
imately at 8 kHz. The peak at 6 kHz is caused by the selfresonance of the earphone.
The effect of the eardrum is clear when comparing
Figs. 19 and 20. The damped eardrum attenuates the peaks
and notches caused by the canal resonances, especially
when measured at the canal entrance. The responses at
the eardrum for different canal lengths are depicted in
Fig. 21. The locations of the first half-wave resonance
peaks, which are marked by arrows, are determined by
the length of the canal. The impedance of the eardrum
affects the response for a large frequency range, whereas
the length of the canal determines the response at frequencies close to 10 kHz.
Fig. 19. Frequency responses from undamped ADECS with 25-mm canal length measured with eardrum microphone (drum) and with
in-ear microphone of EFIM (entrance).
Fig. 20. Frequency responses measured with eardrum microphone (drum) and with in-ear microphone of EFIM (entrance) from
damped ADECS with 25-mm canal length.
Fig. 21. Frequency responses measured with eardrum microphone of damped ADECS (blocked with EFIM) for different canal lengths.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
277
HIIPAKKA ET AL.
PAPERS
3.2.2 Frequency Responses of Human Ears
(Blocked)
The frequency responses of the left ears of eight test
subjects measured with the in-ear microphone of the
EFIM are depicted in Fig. 22. The differences between
the in-ear frequency responses were significant. The
responses below 2 kHz were strongly dependent on the
fitting of the earphone. A loose fitting resulted in leakage
and attenuated low frequencies. In addition the differences
between right and left ears were interestingly large for
some of the subjects. The responses from both ears of one
subject (JV) are depicted in Fig. 23. The lengths of the left
and right canals seem to be different, since the first halfwave resonance peaks are located at 7.3 kHz (right) and
8.3 kHz (left).
The responses at the canal entrance obtained from
real ears showed similarities with those measured from
the ADECS and the DADEC. As one example of this,
Fig. 24 shows the responses measured with the DADEC
(with damped ADECS as ear canal) and the left ear of a
human test subject (MH). The responses are similar up
to 15 kHz. In this example the ear canal length of the
ADECS was set to 17 mm, upon which the artificial ear
adds a few millimeters, as described in Section 1.3 and
Fig. 7.
3.2.3 Measured versus Modeled Frequency
Responses of Blocked ADECS
Having estimated the Thévenin source model, acoustic
responses to any point in the ear canal model can be
Fig. 22. In-ear frequency responses of left ears of eight test subjects measured with EFIM.
Fig. 23. Frequency responses at left and right blocked ear canal entrances of human test subject (JV).
Fig. 24. Frequency responses at blocked ear canal entrances of human test subject (MH) and DADEC.
278
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
computed and compared to measured data. Figs. 25 and 26
show modeled and measured responses at 7 mm from
the earphone (entrance) and at the eardrum (drum) for
two different canal lengths. The resemblance between
modeled and measured data is good, which shows that the
model is applicable when exploring the behavior of the
insert earphone connected to different kinds of ear canals.
4 SUMMARY AND CONCLUSIONS
The aim of this study was to explore the acoustic behavior of the external ear together with insert-type
earphones. Understanding the individual features of listeners and how they affect the earphone-reproduced sound
helps when designing earphones and using them in
binaural reproduction and auralization. Insert earphones
occlude the ear canal so that the effects of concha, pinna,
head, and shoulders are excluded. These external parts
need to be taken into account carefully in detailed
auralization, but in music reproduction their individual
variations are not as prominent.
In addition to the earphone driver itself, the tone color
in insert earphone reproduction is dependent on the acoustic impedance of the eardrum, the size and form of the ear
canal, and the leakage of the earphone fitting. The user
must take great care to obtain tight fitting, because otherwise no full bass response is possible. From our measurement results we conclude that the length of the ear canal
has a clear effect when determining the pressure frequency
responses at the ear canal entrance and at the eardrum. In
addition the eardrum impedance determines the sharpness
of the resonance peaks and antiresonance notches. For a
MODELING OF EXTERNAL EAR ACOUSTICS
realistic model of a human ear canal, the impedance of the
eardrum needs to be taken into account.
Other physical parameters, such as the shape of the
ear canal and other parts of the outer ear were also
studied [30], although they are not reported in this
paper. Differences in shape of the outer ears are factors
to be considered for accurate modeling of the outer ear.
However, compared to the eardrum impedance and ear
canal length, other differences were found to be of less
importance.
The measurement methods and modeling techniques
presented in this paper are applicable when exploring the
behavior of insert earphones coupled to different kinds of
ear canals. Future development of the physical simulators
and the computational model presented include improved
physical and computational models of the eardrum. Direct
measurements of the eardrum pressure on human test subjects during in-ear headphone listening are also of great
interest.
5 ACKNOWLEDGMENT
This work was supported by Nokia Corporation and
project UI-ART at the Helsinki University of Technology.
The authors wish to thank two anonymous reviewers for
their comments.
6 REFERENCES
[1] J. Blauert, Spatial Hearing: The Psychophysics of
Human Sound Localization, rev. ed. (MIT Press, Cambridge, MA, 1997).
Fig. 25. Modeled and measured frequency responses at eardrum and at entrance of undamped ADECS with 26-mm ear canal.
Fig. 26. Modeled and measured frequency responses at eardrum and at entrance of undamped ADECS with 20-mm ear canal.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
279
HIIPAKKA ET AL.
[2] H. Hudde, “Estimation of the Area Function of
Human Ear Canals by Sound Pressure Measurements,”
J. Acoust. Soc. Am., vol. 73, pp. 24–31 (1983).
[3] J. Kates, “A Computer Simulation of Hearing Aid
Response and the Effects of Ear Canal Size,” J. Acoust.
Soc. Am., vol. 83, pp. 1952–1963 (1988 May).
[4] J. P. Dalmont, “Acoustic Impedance Measurement,
Part I: A Review,” J. Sound Vib., vol. 243, pp. 427–439
(2001).
[5] W. Rabinowitz, “Measurement of the Acoustic
Input Immittance of the Human Ear,” J. Acoust. Soc. Am.,
vol. 70, pp. 1025–1036 (1981).
[6] D. Keefe, R. Ling, and J. Bulen, “Method to Measure Acoustic Impedance and Reflection Coefficient,”
J. Acoust. Soc. Am., vol. 91, pp. 470–485 (1992).
[7] H. Hudde, “Measurement of the Eardrum Impedance of Human Ears,” J. Acoust. Soc. Am., vol. 73, pp.
242–247 (1983).
[8] J. B. Allen, “Measurement of Eardrum Acoustic Impedance,” in Peripheral Auditory Mechanisms, J. B. Allen,
J. L. Hall, A. Hubbard, S. T. Neely, and A. Tubis, Eds.
(Springer, New York, 1986), pp. 44–51.
[9] J. Chan and C. Geisler, “Estimation of Eardrum
Acoustic Pressure and of Ear Canal Length from Remote
Points in the Canal,” J. Acoust. Soc. Am., vol. 87,
pp. 1237–1247 (1990).
[10] D. Keefe, J. Bulen, K. Arehart, and E. Burns, “EarCanal Impedance and Reflection Coefficient in Human
Infants and Adults,” J. Acoust. Soc. Am., vol. 94, pp.
2617–2638 (1993 Nov.).
[11] S. Voss and J. Allen, “Measurement of Acoustic
Impedance and Reflectance in the Human Ear Canal,”
J. Acoust. Soc. Am., vol. 95, pp. 372–384 (1994).
[12] H. Hudde, A. Lodwig, and A. Engel, “A WideBand Precision Acoustic Measuring Head,” Acustica, vol.
82, pp. 895–904 (1996).
[13] H. Hudde, A. Engel, and A. Lodwig, “Methods for
Estimating the Sound Pressure at the Eardrum,” J. Acoust.
Soc. Am., vol. 106, pp. 1977–1992 (1999).
[14] S. Voss, J. Rosowski, C. Shera, and W. Peake,
“Acoustic Mechanisms that Determine the Ear-Canal
Sound Pressures Generated by Earphones,” J. Acoust.
Soc. Am., vol. 107, pp. 1548–1565 (2000).
[15] J. P. Dalmont, “Acoustic Impedance Measurement,
Part II: A New Calibration Method,” J. Sound Vib., vol.
243, pp. 441–459 (2001).
[16] B. Farmer-Fedor and R. Rabbitt, “Acoustic Intensity, Impedance and Reflection Coefficient in the Human
Ear Canal,” J. Acoust. Soc. Am., vol. 112, pp. 600–620
(2002).
[17] D. Ćirić and D. Hammershøi, “Acoustic Impedances of Ear Canals Measured by Impedance Tube,” in
Proc. Int. Cong. on Acoustics (Madrid, Spain, 2007 Sept.),
pp. 1–6.
[18] D. Ćirić and D. Hammershøi, “Coupling of
Earphones to Human Ears and to Standard Coupler,”
J. Acoust. Soc. Am. (2006 Oct.).
[19] H. Møller, D. Hammershøi, C. B. Jensen, and
M. F. Sørensen, “Transfer Characteristics of Headphones
280
PAPERS
Measured on Human Ears,” J. Audio Eng. Soc., vol. 43,
pp. 203–217 (1995 Apr.).
[20] D. Hammershøi and H. Møller, “Sound Transmission to and within the Human Ear Canal,” J. Acoust. Soc.
Am., vol. 120, pp. 2096–2107 (1996 Jan.).
[21] ANSI S3.25-1989, “American National Standard
for an Occluded Ear Simulator,” American National Standards Institute, New York (1989).
[22] IEC 60711, “Occluded-Ear Simulator for the Measurement of Earphones Coupled to the Ear by Ear Inserts,”
International Electrotechnical Commission, Geneva, Switzerland (1981 Jan.).
[23] ITU-T P.57, “Telephone Transmission Quality,
Telephone Installation, Local Line Networks—Objective
Measuring Apparatus—Artificial Ears,” International
Telecommunications Union, Geneva, Switzerland (2005
Nov.).
[24] S. Paul, “Binaural Recording Technology: A Historical Review and Possible Future Developments,” Acta
Acustica Acustica, vol. 95, pp. 1–22 (2009 Aug.).
[25] F. Firestone, “The Phase Difference and Amplitude
Ratio at the Ears due to a Source of Pure Tone,” J. Acoust.
Soc. Am., vol. 2, pp. 260–270 (1930 Jan.).
[26] A. Farina, “Simultaneous Measurement of Impulse
Response and Distortion with a Swept-Sine Technique,”
presented at the 108th Convention of the Audio Engineering Society, J. Audio Eng. Soc. (Abstracts), vol. 48, p. 350
(2000 Apr.), preprint 5093.
[27] M. Stinson and B. Lawton, “Specification of the
Geometry of the Human Ear Canal for the Prediction
of Sound-Pressure Level Distribution,” J. Acoust. Soc.
Am., vol. 85, pp. 2492–2503 (1989 Jan.).
[28] J. Fay, S. Puria, and C. Steele, “The Discordant
Eardrum,” Proc. Nat. Acad. Sci., vol. 103 (2006 Dec.).
[29] M. Stinson, “The Spatial Distribution of
Sound Pressure within Scaled Replicas of the Human
Ear Canal,” J. Acoust. Soc. Am., vol. 78, pp. 1596–1602
(1985 Jan.).
[30] M. Hiipakka, “Measurement Apparatus and
Modelling Techniques of Ear Canal Acoustics,” Master’s
thesis, Helsinki University of Technology, Espoo, Finland
(2008 Dec.).
[31] D. Raichel, The Science and Applications of
Acoustics (Springer, New York, 2000).
[32] L. L. Beranek, Acoustics 3rd printing, (Acoustical
Society of America, 1988).
[33] P. M. Morse and K. U. Ingard, Theoretical
Acoustics (Princeton University Press, Princeton, NJ,
1986).
[34] C. Poldy, “Headphones,” in Loudspeaker and
Headphone Handbook, J. Borwick, Ed., 3rd ed. (Focal
Press, Woburn, MA, 2001), chap. 14.
[35] C. Poldy, “Headphone Fundamentals,” Tutorial
presented at the 120th Convention of the Audio Engineering Society, Paris, 2006 May, http://www.aes.org/
tutorials/.
[36] U. Ingård, “On the Radiation of Sound into a
Circular Tube, with an Application to Resonator,”
J. Acoust. Soc. Am., vol. 20, pp. 665–682 (1948).
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
PAPERS
MODELING OF EXTERNAL EAR ACOUSTICS
THE AUTHORS
M. Hiipakka
M. Tikander
Marko Hiipakka was born in Espoo, Finland, in 1973.
He received an M.Sc. (Eng.) degree from the Helsinki
University of Technology in 2008, majoring in acoustics
and audio signal processing.
He joined the Laboratory of Acoustics and Audio Signal Processing in 2008 May and was named a researcher
in 2009 January. Working with physical simulators, computational modeling, and particle velocity transducers as
his essential tools, his current research and postgraduate
studies focus on the acoustics of the outer ear, headphones, and spatial sound.
Mr. Hiipakka is the secretary of the Acoustical Society
of Finland.
l
Miikka Tikander studied electrical and communication
engineering at the Helsinki University of Technology,
Espoo, Finland. After graduating in acoustics and audio
signal processing, he continued as a researcher in the Laboratory of Acoustics and Audio Signal Processing at the
Helsinki University of Technology. He received a Ph.D.
degree in acoustics and audio signal processing in 2009.
His research work included topics such as acoustic positioning, headset modeling, and augmented reality audio.
In 2009 he joined Nokia Gear, where he works as an
audio and acoustics specialist.
Dr. Tikander is a board member of the Acoustical Society of Finland and the Finnish Section of the Audio Engineering Society.
J. Audio Eng. Soc., Vol. 58, No. 4, 2010 April
M. Karjalainen
Matti Karjalainen was born in Hankasalmi, Finland,
in 1946. He received M.Sc. and Dr.Sc. (Tech.) degrees
in electrical engineering from the Tampere University
of Technology, Tampere, Finland, in 1970 and 1978,
respectively.
He is a professor at the Helsinki University of Technology, Espoo, Finland, and founded audio signal processing education there in the 1980s. His main interest is
in audio signal processing, such as DSP for sound
reproduction and auralization, music DSP, and sound
synthesis, as well as perceptually based signal processing. In addition his research activities cover speech
processing, perceptual auditory modeling, spatial hearing, DSP hardware, software, and programming environments, as well as various branches of acoustics,
including musical acoustics and modeling of musical
instruments.
Dr. Karjalainen has written 380 scientific and engineering papers and contributed to organizing several conferences and workshops, such as serving as the papers
chairman of the AES 16th International Conference and
as technical chairman of the AES 22nd International
Conference. He is a Fellow and silver medalist of the
Audio Engineering Society, a Fellow of the Institute of
Electrical and Electronics Engineers, and a member of the
Acoustical Society of America, the European Acoustics
Association, the International Speech Communication
Association, and several Finnish scientific and engineering societies.
281